In this work, a numerically designed surface plasmon resonance (SPR) sensor is proposed for non-invasive detection of glucose concentration in human urine, with potential applicability toward future experimental realization. The sensor performance is investigated using a simulated Au/Si \(_3\) N \(_4\) bilayer structure with optimized thicknesses of 50/12 nm. Glucose concentrations ranging from 0.625 to 10 g/dL are analyzed by monitoring resonance wavelength shifts induced by refractive index variations. The numerical results indicate a systematic resonance shift from 1010 to 1160 nm, accompanied by a maximum confinement loss of 292 dB cm−1. A peak wavelength sensitivity of 12,500 nm RIU−1 is achieved at a glucose concentration of 10 g/dL, while a sensitivity of approximately 10,000 nm RIU−1 is maintained at lower concentrations. Furthermore, the signal-to-noise-ratio-dependent detection limit is found to vary between 31.5 and 56 nm, with accuracy values ranging from 0.00705 to 0.01145 nm \(^{-1}\) . Although the present study is entirely simulation-based, the obtained results highlight the strong potential of the proposed SPR sensor architecture for high-sensitivity glucose monitoring, motivating future fabrication and experimental validation for real-time clinical applications.
